|
 |
 |
|
Indexing |
The Indexing Process at BIMThe important thing to mention about the indexing process at BIM is that our indexes are never computer-generated. Computer-generating the index might seem like the fast and easy way to produce an index, however the results leave much to be desired. The reason for this is the way that computer programs generate indexes. Simply put, they create a list of frequently occurring words. In essence, what is created is actually a concordance, not an index. Publication readers or Web site visitors that use the computer- generated concordance would be directed to every occurrence of each word within the index. Even if just passing mention is made of the topic, with no "real" information, the user would be led to those pages. Thus, the concordance would be practically useless. Also, computers do not have the ability to scan text and pick out concepts that are presented within the text even if the text does not state the concept directly. In addition, it does not take into consideration that the searcher may be thinking of a different word than is in the text, although the word he is thinking of correctly conveys the concept. Does this mean that the individual doing the indexing reads through every bit of material on a site, or in a publication? It means precisely that. And she reads it as no one else ever will, examining how to present the information in the index, what words the searcher might use instead of the ones in the text, and what types of cross-references should be made. A professional indexer is trained in, and has experience in how to manage information. UPDATE: I've recently been contacted by three separate companies, each of which claims to have a Web site search system which can do what we just described above, thus imitating human intellect in recognizing relevant information. None of the companies have yet been able to give me a demonstration of their product, although one plans on doing so in mid-2002. Even at that, the price for the system will start at $90,000 US. I cannot give my viewpoint yet on the system as I haven't yet been able to evaluate it. However, if it does work as they profess it to work, the price most certainly makes it prohibitive or at least an unwise purchase for a smaller Web site of between 100 to 5000 pages. It would probably be best used for extremely large Web sites and/or intranets having between 5000 to a million or so pages. Sites with less pages than that would have at least as good or better of a system for substantially less money using human indexers. Have more questions about how we index? Call 1-877-205-9259 or e-mail us at broccoli@bim.net . |